Back to all
At a glance:
- 53% of shoppers trust AI tools, including an AI shopping assistant, as much as brand websites, according to a Rithum and Retail Dive survey. This trust is reshaping AI shopping verification behavior and expectations for AI in retail.
- When shoppers verify an AI recommendation, retailer and brand sites rank near the bottom at 5%, showing how AI shopping verification is happening away from owned channels.
- 64% of shoppers ages 18 to 27 say they’re likely to purchase based on an AI recommendation without verifying it anywhere else.
- AI-referred visitors now convert 42% higher than non-AI traffic.
- Retailers like Walmart, Target, and Instacart are enabling purchases directly inside AI conversations, accelerating a future where the entire shopping journey happens in one place.
A year ago, shoppers arriving through AI tools browsed but left without buying. A year later, those same shoppers are 42% more likely to buy than shoppers arriving through traditional channels1. In the same month, Walmart deployed its AI shopping agent inside ChatGPT2, joining Target, Instacart, and DoorDash in letting shoppers browse, compare, and buy products directly inside the conversation.
In Rithum and Retail Dive’s survey of 1,046 online shoppers across the U.S. and U.K., 53% already trust AI tools as much as brand websites. And when they want a second opinion on what AI told them, they’re going everywhere except the brand to get it.
When shoppers double-check, they go everywhere but the brand site
When shoppers verify an AI recommendation, they’re choosing channels outside the brand’s control. Search engines are the top destination at 28%. Online reviews come next at 19%, followed by friends and family at 17%. Retailer and brand sites rank near the bottom at 5%.
A brand’s own product page, no matter how thorough, is still the brand talking about itself. Shoppers want independent voices, and they’re finding them everywhere else.
That puts more pressure on the product data traveling through those channels. If the search engine is the second stop after AI, the data you’re pushing into Google, Bing, and other platforms needs to be accurate and complete. If a shopper pulls up a review site and finds specs that conflict with what the AI told them, the brand absorbs that cost. In the same survey, 58% of shoppers said trust in the brand decreases when an AI recommendation contains incorrect product information, and 16% abandon the purchase entirely.
Brands have the answers, but shoppers are asking somewhere else
A shopper asks an AI tool to recommend a running shoe for flat feet under $150. Three options come back. The shopper likes one but wants to confirm the arch support claim before buying.
They type the product name into Google. They scan a couple of review sites. They text a friend who runs. The brand’s product page may have the most detailed answer to their question, but the shopper has already moved on to other sources.
Product information accuracy across your entire distribution footprint now carries more weight than the quality of your own site experience. Feeds, marketplace listings, third-party retailer pages, and structured data that AI tools can parse all shape what the shopper encounters during verification. The brands investing in that full footprint are the ones staying in the consideration set. The ones focused primarily on their own site are building for a shopping journey that fewer customers follow.
A growing share of shoppers skip verification entirely
Among shoppers ages 18 to 27, 64% say they’re likely to purchase based on an AI recommendation without verifying it anywhere else. Higher-income shoppers are twice as likely to trust AI without visiting another site. And across all demographics, 32% say they spend less time browsing other sites after using an LLM.
For these buyers, the AI recommendation from an AI shopping assistant is the decision. The brand site is largely absent from it. And shoppers who verify through search and reviews encounter a broader set of options than they would on a single brand’s site, giving unfamiliar brands a real opening to enter the consideration set with accurate, well-distributed product data. Shoppers who skip verification altogether are relying entirely on whatever the AI already knows about your product.
The verification step itself is disappearing
64% of shoppers already take AI at its word. The platforms coming next are built to make that feel even more natural.
AI agents could mediate $3 trillion to $5 trillion of global consumer commerce by 2030, according to McKinsey and Co. Two competing open protocols are already live and processing transactions end to end: OpenAI and Stripe’s Agentic Commerce Protocol and Google’s Universal Commerce Protocol. The AI agent handles product discovery, comparison, checkout in one place.
Now picture that same running shoe shopper six months from now. They ask ChatGPT the same question. A product card appears with an image, price, and a “Buy” button. They tap it, confirm their saved payment method, and the order ships. The entire transaction happened inside a single conversation.
The shoppers who still verify aren’t going back to the brand to do it. They’re checking reviews, search results, other people. And as AI agents take on more of that process, the brand’s window to influence the answer gets smaller. The data has to be right before the question is ever asked.”
Your product data is now your pitch to an AI buyer that will never visit your homepage
Getting product data right across every channel is the minimum. It’s expected. The question is where that data lives: search engines, review platforms, marketplace listings, and the structured data feeds that AI agents read when they decide what to recommend. AI-readable product content needs to be complete, consistent, and built for machines to parse. For a growing number of shoppers, that content is the only version of your brand they’ll ever see.
The distance between discovery and purchase is collapsing. Sometimes it’s a single conversation with an AI agent. The brands feeding that conversation with accurate, well-distributed product data are the ones the agent recommends.
For a full breakdown of the data, download The New Discovery Engine report.
Sources:
1: https://www.retailtouchpoints.com/features/the-agentic-commerce-paradox-its-already-here-and-its-also-still-evolving/618945/
2: https://www.cbsnews.com/news/ai-agentic-artificial-inteligence-what-is-it/
